摘要
近年来我国创业板股市频繁出现新股破发现象,暴露出创业板市场存在的风险问题。基于行为金融学及有限注意力理论,运用Web挖掘手段和机器学习算法分析股票论坛投资者的文本评论和搜索行为,建立投资者情绪和投资者关注指数,对创业板新股破发进行定量化实证研究。结果表明,除了市场指标、发行指标、机构参与指标和财务指标,从股票论坛和搜索引擎获取的投资者情绪和关注也是影响创业板股票破发的重要因素,据此建立的新股破发预测模型平均准确率达90%。
In recent years, lots of new shares in GEM break on the first trading day, which shows the inefficiency of the IPO pricing in GEM. Based on behavioral finance and limited attention theory, we analyze investors' online review and search queries with Web mining technique, then do empirical study on the determinants for IPO pricing in Chinese GEM from the perspective of investors' sentiment and attention. The result shows that apart from traditional financial factors, investors' sentiment and attention indexes are also important factors influencing IPO underpricing. The accuracy of the prediction model reaches 90%.
出处
《微型机与应用》
2015年第10期58-60,共3页
Microcomputer & Its Applications
关键词
WEB挖掘
新股破发
机器学习
支持向量机
朴素贝叶斯
Web mining
IPO underpricing
machine learning
support vector machine(SVM)
naive Bayes(NB)